The Observability Cost Problem
When you start building production systems, observability feels like a solved problem. You reach for Datadog, New Relic, or Elastic, configure a few agents, and logs and traces start flowing. Then the invoice arrives.
Modern observability platforms are priced on data ingestion volume — and in a world where microservices, Kubernetes, and distributed architectures generate petabytes of telemetry per month, the cost scales exponentially. Teams with 20 services routinely see $20,000–$80,000 monthly bills just for log management.
This is the problem we set out to solve at XplurData: build an open source, free observability platform that gives engineering teams the same power as commercial solutions — without the pricing shock.
Why OpenTelemetry (OTel)?
OpenTelemetry (OTel) is the CNCF standard for collecting telemetry signals — logs, metrics, and traces — from any application, in any language, on any infrastructure. It is vendor-neutral, widely adopted, and actively maintained by contributions from Google, Microsoft, AWS, Honeycomb, and hundreds more.
The core of the OTel ecosystem is the OpenTelemetry Collector: a pipeline daemon that receives telemetry over OTLP (gRPC or HTTP), applies transformations, and exports it to any backend. Most commercial platforms require proprietary agents that lock you to their ecosystem. OTel breaks that lock.
The OTel Collector Pipeline
In XplurData, the OTel Collector is the data gateway. A typical otel-config.yaml pipeline looks
like:
receivers:
otlp:
protocols:
grpc:
endpoint: 0.0.0.0:4317
http:
endpoint: 0.0.0.0:4318
processors:
batch:
timeout: 5s
send_batch_size: 1000
exporters:
otlphttp:
endpoint: http://xd-api:8080/v1/otlp
compression: gzip
service:
pipelines:
logs:
receivers: [otlp]
processors: [batch]
exporters: [otlphttp]
traces:
receivers: [otlp]
processors: [batch]
exporters: [otlphttp]This single config file connects any OTel-instrumented application directly to the XplurData API. No proprietary SDK, no data plane agent, no vendor lock-in.
Why Apache Doris?
Choosing the right analytics backend is the hardest part of building a high-performance observability tool. We evaluated several candidates:
| Database | Query Latency | Full-Text Search | Compression | Self-Hosted |
|---|---|---|---|---|
| Apache Doris | ✅ Sub-second | ✅ Native BM25 | ✅ High (columnar) | ✅ Yes |
| Elasticsearch | ⚠️ Moderate | ✅ Native Lucene | ⚠️ Moderate | ✅ Yes |
| ClickHouse | ✅ Sub-second | ⚠️ Limited | ✅ High | ✅ Yes |
| Loki | ⚠️ Index-free only | ❌ No | ✅ High | ✅ Yes |
Apache Doris stood out for several reasons:
-
Native BM25 full-text search — Doris supports full-text search directly in SQL using
MATCH,MATCH_ALL,MATCH_ANY,MATCH_PHRASE, and BM25 relevance scoring — no separate search cluster needed. - MPP architecture — Massively Parallel Processing allows Doris to fan out queries across Backend nodes, achieving sub-second response even on multi-billion-row tables.
- Columnar storage — OTel log entries are wide (many fields, most sparse). Columnar storage with high compression means storing 1 TB of raw logs in as little as 100 GB on disk.
- MySQL-compatible SQL interface — Engineers already know SQL. No new query language to learn.
- Fully open source — Apache License 2.0. No cloud-only features, no enterprise-only editions.
The XplurData Platform Architecture
XplurData is composed of four independent, containerized services that work together:
Every component is containerized. The entire stack starts with a single install script:
bash -c "$(curl -fsSL \
https://raw.githubusercontent.com/xplurdata/oss-stack/main/install.sh)"What "Free" Actually Means
When we say XplurData is free, we mean:
- Zero license fees — AGPLv3 means you can run it commercially, modify it, and distribute it, as long as you share improvements.
- Zero per-GB pricing — You ingest as much data as your Doris cluster can store. Scale up by adding more Doris Backend nodes.
- Zero vendor lock-in — Your data lives in Apache Doris tables you control. Export it to any format at any time.
- You pay only infrastructure — Run it on a $20/month VPS or on a 50-node Kubernetes cluster. The software cost is always $0.
Real-World Performance Numbers
In our internal testing with a 3-node Doris cluster (3× Doris FE, 3× Doris BE):
Getting Started
You can have the full XplurData stack running in under 5 minutes on any Linux or macOS machine with Docker:
- Run the one-command installer:
bash -c "$(curl -fsSL ...install.sh)" - Generate test telemetry with
telemetrygenor instrument your app with any OTel SDK - Open
http://localhost:3000and explore your logs, traces, and metrics
Conclusion
The observability market doesn't have to be expensive. By combining the vendor-neutral OpenTelemetry standard with the raw analytical power of Apache Doris, XplurData delivers a genuinely free, genuinely scalable, and genuinely open alternative to commercial platforms.
If you're tired of surprise observability bills, give XplurData a try. It costs nothing but a few minutes to set up.
Ready to Explore XplurData?
Deploy the platform, connect your applications using OpenTelemetry and start exploring logs and traces with zero software licensing costs.
Download XplurData